generating a boot tape and make_net_recovery

[Jose Javier Sianes Ruiz]

I'm trying to make a bootable tape like described in makenetrec.txt text document from a previusly make_net_recovery image. I have made the image using Ignite UI with no error or warnnings, but when I make dd to create the tape, like described on ' Step 2 - Write the tape', I allways get this error:

root@admorum:/var/opt/ignite/clients/klax/recovery/latest> es/klax/2005-03-10,11:37 of=/dev/rmt/0mn obs=10k <
I/O error
3205820+0 records in
160290+1 records out
root@admorum:/var/opt/ignite/clients/klax/recovery/latest>


Anyone can tell me a way to verify if the tape is correctly generated (witout rebooting the machine) and tell me why this I/O error could be happend?. Thanks.

[Steven E. Protter]

This tape is almost certainly not good. I would not rely on a tape created with an i/o error.

I'm wondering why we don't just do a make_tape_recovery. I'm guessing that is becasue this system does not have a tape drive.

SEP

[Florian Heigl (new acc)]

the last record wasn't properly written, so the only way to safely compare would be re-reading and checksumming the image from tape.
if the image was 160290 or 160291 10k blocks in size, things might be okay, but thinking technically the tape is definitely *not*, as it's different from what You had on Your disk.

You said 'always' - did You try different tapes? In that case this might be a blocksize problem.

(SEP,
imagine he recovered a two-months old image from a backup of /var/opt/ignite/recovery/archives/ - we also have this situation from time to time.)

[Jose Javier Sianes Ruiz]

Yes, this is the second tape I used with the same result in the same point. If you have make some tape this way, you have to make 3 points: first, rew the tape; second, dd lif file at the begining of the tape with 2k block size and third (and last), dd with 10k the system image generated by ignite. In this last point it when the process fails with I/O error. Any sugestion to solve the problem, maybe changing block size?


PS: The reason I made this process is because I have a Ignite-UX server who store in disk several sysimages of some HP-UX servers. I only generate a tape when its needed (for example system recovery). Its an affective way to centralize whole process and reduce hand work.

[Mel Burslan]

If the machine used to create the tapes has a second tape drive, or have a possibility of attaching one, I would try this first. Two tapes failing with the same problem is a little too much of coincidence for me. You may have a bad tape drive.

[Jose Javier Sianes Ruiz]

The tape unit is good, I think is more probably a cartridge problem. I have passed a cleaning cartridge to the unit and I'm rewriting the tape with a 2K block size on system image. When its finish (more than an our ) I will explain the results. Thanks for your help.

[Jose Javier Sianes Ruiz]

Im using HP gunzip with pax patches and the command suggested:

# cat 2005-03-10,11:37 | gunzip | pax -tf - | head -5

Works perfectly, so it cuould be a gunzip problem in my case. My image file is 2.5GB large and with two different tapes (and passing a cleanning cartridge between) it allways fails in the same record:

root@igniteUXServer:/root> ll ./2005-03-10,11:37
-rw------- 1 bin sys 2567972913 Mar 10 12:38 ./2005-03-10,11:37
root@igniteUXServer:/root> cat 2005-03-10,11:37 | gunzip | pax -tf - | head -5
stand/
stand/lost+found/
stand/ioconfig
stand/bootconf
stand/system
root@igniteUXServer:/root>
root@igniteUXServer:/root>dd if=./2005-03-10,11:37 of=/dev/rmt/0mn obs=10k
I/O error
3190420+0 records in
159520+1 records out
root@igniteUXServer:/root>

Also, making a dd to /dev/null works perfectly:

root@igniteUXServer:/root> dd if=./2005-03-10,11:37 of=/dev/null obs=10k
5015572+1 records in
250778+1 records out
root@igniteUXServer:/root>

Help will be very apreciated. Thanks.

[Nick W]

Jose,

Before you blame the target device (tape-drive or suspect media) check that the archive image you have is not corrupted (as far as gzip is concerned - dd will transfer the data as it does not care what format it is in :)

cat 2005-03-10,11:37 | gunzip | pax -tf -

ie let it run all the way through.

Is the tape-media capable of storing the full (uncompressed...?) image - hitting the end-of-media could be construed as an IO error....? Does the fact that your image file is >2GB pose any problems - perhaps someone else knows the answer to these ....?



HTH
Regards,
Nick

[Jose Javier Sianes Ruiz]

Nick, the command you say works perfectly, showing all files includes in ignite imagefile. The errorlevel after execute it is 0. Im going to change present tape for a DDS3, and try again. More news in over two hours :)

[Jose Javier Sianes Ruiz]

The DDS2 tape using the first and the second time was corrupt!!!! I have solve it changing tape for a new DDS3 and all goes right now:

root@igniteUXServer:/root> dd if=/var/opt/ignite/recovery/archives/klax/2005-03-10,11:37 of=/dev/rmt/0mn obs=10k
5015572+1 records in
250778+1 records out
root@igniteUXServer:/root>

Thanks everybody for your help.